4 ms·

Despite having thought of the dangers of AI, this literally never occurred to me. Humans don't create AI, machines do.

by phaemon 10y ago

LesZedCB 10y ago
Imagine if we had the unambiguous source code for our own bodies? AI will be able to have that without any work, and can modify and recompile with little to no effect and all benefit in the matter of seconds, where humans have to wait years. This is referred to as the intelligence explosion[0] or AI Foom[1]. [0] https://en.wikipedia.org/wiki/Intelligence_explosion https://en.wikipedia.org/wiki/Intelligence_explosion [1] https://wiki.lesswrong.com/wiki/The_Hanson-Yudkowsky_AI-Foom_Debate https://wiki.lesswrong.com/wiki/The_Hanson-Yudkowsky_AI-Foom...
- mastazi 10y ago
  > Imagine if we had the unambiguous source code for our own bodies? We sequenced human DNA, so I think we already have something kinda like it. I'm not a scientist but my basic understanding is that while we have the whole sequence (in your source code analogy, we can do git clone or git pull) we are not really good at tweaking it ("oops this codebase uses architectural solutions we're not familiar with, we will have to study it extensively before we can deploy those proposed changes to the production environment")
  - jcranmer 10y ago
    DNA is not really source code. At best, it's kind of like... an algorithm textbook. Only a very tiny portion of DNA (like, ~1-2% in Homo sapiens) codes for proteins in the manner that you would have learned in biology class. A much larger fraction is used somehow, but how much and what its functions are is still a matter of major debate. The effort to sequence the human genome has also led us to discover that there are other ways in which traits can be heritable that don't involve changes to DNA. The focus now is largely on epigenetics, although explanations invoking microbiomes of bacteria inside our organs was definitely popular for a while. In short, we really don't know a whole lot about our own molecular biology, and a lot of the research in the past 60 years since the discovery of DNA has tended to show "there's more going on than we thought." Where things involve just DNA, we have very good tools for reading (sequencing) and writing (CRISPR/Cas9) it. What we don't have good tools for is modifying our epigenetics, and we don't have a good handle on what comes from DNA and what comes from epigenetics.
    - mastazi 10y ago
      Thanks, I'm interested in this topic, would you be able to suggest sources that are understandable by a layperson?
      - jcranmer 10y ago
        Unfortunately, I'm no expert in molecular biology. Some of the more recent stuff I've picked up by following Derek Lowe's blog (http://blogs.sciencemag.org/pipeline/ http://blogs.sciencemag.org/pipeline/), which occasionally discusses some new results in molecular biology.
      - allenz 10y ago
        The standard way to learn this stuff is a course/textbook followed by reading the important papers. I recommend https://ocw.mit.edu/courses/biology/7-28-molecular-biology-spring-2005/ https://ocw.mit.edu/courses/biology/7-28-molecular-biology-s...: it is well-organized and a great use of time, even though it doesn't get into the latest most exciting topics. If you just want an overview of what's going on, I curated these links for you. For each, focus on the main idea and why it is significant for accomplishing biological goals. https://en.wikibooks.org/wiki/Human_Physiology/Homeostasis#Overview https://en.wikibooks.org/wiki/Human_Physiology/Homeostasis#O... https://en.wikipedia.org/wiki/Molecular_biology https://en.wikipedia.org/wiki/Molecular_biology section 1-2 Skim https://en.wikipedia.org/wiki/Central_dogma_of_molecular_biology https://en.wikipedia.org/wiki/Central_dogma_of_molecular_bio... https://www.khanacademy.org/science/biology/classical-genetics/molecular-basis-of-genetics-tutorial/ https://www.khanacademy.org/science/biology/classical-geneti... watch 2x speed https://en.wikipedia.org/wiki/Genetics#Research_methods https://en.wikipedia.org/wiki/Genetics#Research_methods https://en.wikipedia.org/wiki/Cellular_communication_(biology) https://en.wikipedia.org/wiki/Cellular_communication_(biolog... https://www.khanacademy.org/science/biology/cell-signaling https://www.khanacademy.org/science/biology/cell-signaling https://mcb.berkeley.edu/courses/mcb110spring/nogales/mcb110_s2008_4signaling.pdf https://mcb.berkeley.edu/courses/mcb110spring/nogales/mcb110... https://en.wikipedia.org/wiki/Epigenetics https://en.wikipedia.org/wiki/Epigenetics intro and diagram only http://www.zymoresearch.com/learning-center/epigenetics/what-is-epigenetics http://www.zymoresearch.com/learning-center/epigenetics/what... all pages https://www.jove.com/science-education-database/2/basic-methods-in-cellular-and-molecular-biology https://www.jove.com/science-education-database/2/basic-meth... understand basic research methods https://www.khanacademy.org/science/biology/human-biology#immunology https://www.khanacademy.org/science/biology/human-biology#im... skim https://en.wikipedia.org/wiki/Induced_pluripotent_stem_cell https://en.wikipedia.org/wiki/Induced_pluripotent_stem_cell https://www.cancerquest.org/cancer-biology/ https://www.cancerquest.org/cancer-biology/ is awesome. I recommend angiogenesis, metastasis, and tumor-host interactions
Capt-RogerOver 10y ago
What other dangers did you think of? It seem the only dangers all stem exactly from this: when the technology becomes exponential (machines making machines), at which point they become better at it than humans, and humans can't longer predict the outcome. The technological singularity. The black hole and the big bang of the conscious technology, with results that we can't begin to fathom how to fathom. It's all based on the assumption that AI will make other better AI. It's not an option, it's an inevitable and logical progression for ML tech.

Despite having thought of the dangers of AI, this literally never occurred to me. Humans don't create AI, machines do.

by phaemon 10y ago

Despite having thought of the dangers of AI, this literally never occurred to me. Humans don't create AI, machines do.

Imagine if we had the unambiguous source code for our own bodies? AI will be able to have that without any work, and can modify and recompile with little to no effect and all benefit in the matter of seconds, where humans have to wait years. This is referred to as the intelligence explosion[0] or AI Foom[1]. [0] https://en.wikipedia.org/wiki/Intelligence_explosion https://en.wikipedia.org/wiki/Intelligence_explosion [1] https://wiki.lesswrong.com/wiki/The_Hanson-Yudkowsky_AI-Foom_Debate https://wiki.lesswrong.com/wiki/The_Hanson-Yudkowsky_AI-Foom...

mastazi 10y ago
> Imagine if we had the unambiguous source code for our own bodies? We sequenced human DNA, so I think we already have something kinda like it. I'm not a scientist but my basic understanding is that while we have the whole sequence (in your source code analogy, we can do git clone or git pull) we are not really good at tweaking it ("oops this codebase uses architectural solutions we're not familiar with, we will have to study it extensively before we can deploy those proposed changes to the production environment")
- jcranmer 10y ago
  DNA is not really source code. At best, it's kind of like... an algorithm textbook. Only a very tiny portion of DNA (like, ~1-2% in Homo sapiens) codes for proteins in the manner that you would have learned in biology class. A much larger fraction is used somehow, but how much and what its functions are is still a matter of major debate. The effort to sequence the human genome has also led us to discover that there are other ways in which traits can be heritable that don't involve changes to DNA. The focus now is largely on epigenetics, although explanations invoking microbiomes of bacteria inside our organs was definitely popular for a while. In short, we really don't know a whole lot about our own molecular biology, and a lot of the research in the past 60 years since the discovery of DNA has tended to show "there's more going on than we thought." Where things involve just DNA, we have very good tools for reading (sequencing) and writing (CRISPR/Cas9) it. What we don't have good tools for is modifying our epigenetics, and we don't have a good handle on what comes from DNA and what comes from epigenetics.
  - mastazi 10y ago
    Thanks, I'm interested in this topic, would you be able to suggest sources that are understandable by a layperson?
    - jcranmer 10y ago
      Unfortunately, I'm no expert in molecular biology. Some of the more recent stuff I've picked up by following Derek Lowe's blog (http://blogs.sciencemag.org/pipeline/ http://blogs.sciencemag.org/pipeline/), which occasionally discusses some new results in molecular biology.
    - allenz 10y ago
      The standard way to learn this stuff is a course/textbook followed by reading the important papers. I recommend https://ocw.mit.edu/courses/biology/7-28-molecular-biology-spring-2005/ https://ocw.mit.edu/courses/biology/7-28-molecular-biology-s...: it is well-organized and a great use of time, even though it doesn't get into the latest most exciting topics. If you just want an overview of what's going on, I curated these links for you. For each, focus on the main idea and why it is significant for accomplishing biological goals. https://en.wikibooks.org/wiki/Human_Physiology/Homeostasis#Overview https://en.wikibooks.org/wiki/Human_Physiology/Homeostasis#O... https://en.wikipedia.org/wiki/Molecular_biology https://en.wikipedia.org/wiki/Molecular_biology section 1-2 Skim https://en.wikipedia.org/wiki/Central_dogma_of_molecular_biology https://en.wikipedia.org/wiki/Central_dogma_of_molecular_bio... https://www.khanacademy.org/science/biology/classical-genetics/molecular-basis-of-genetics-tutorial/ https://www.khanacademy.org/science/biology/classical-geneti... watch 2x speed https://en.wikipedia.org/wiki/Genetics#Research_methods https://en.wikipedia.org/wiki/Genetics#Research_methods https://en.wikipedia.org/wiki/Cellular_communication_(biology) https://en.wikipedia.org/wiki/Cellular_communication_(biolog... https://www.khanacademy.org/science/biology/cell-signaling https://www.khanacademy.org/science/biology/cell-signaling https://mcb.berkeley.edu/courses/mcb110spring/nogales/mcb110_s2008_4signaling.pdf https://mcb.berkeley.edu/courses/mcb110spring/nogales/mcb110... https://en.wikipedia.org/wiki/Epigenetics https://en.wikipedia.org/wiki/Epigenetics intro and diagram only http://www.zymoresearch.com/learning-center/epigenetics/what-is-epigenetics http://www.zymoresearch.com/learning-center/epigenetics/what... all pages https://www.jove.com/science-education-database/2/basic-methods-in-cellular-and-molecular-biology https://www.jove.com/science-education-database/2/basic-meth... understand basic research methods https://www.khanacademy.org/science/biology/human-biology#immunology https://www.khanacademy.org/science/biology/human-biology#im... skim https://en.wikipedia.org/wiki/Induced_pluripotent_stem_cell https://en.wikipedia.org/wiki/Induced_pluripotent_stem_cell https://www.cancerquest.org/cancer-biology/ https://www.cancerquest.org/cancer-biology/ is awesome. I recommend angiogenesis, metastasis, and tumor-host interactions

Capt-RogerOver 10y ago

What other dangers did you think of? It seem the only dangers all stem exactly from this: when the technology becomes exponential (machines making machines), at which point they become better at it than humans, and humans can't longer predict the outcome. The technological singularity. The black hole and the big bang of the conscious technology, with results that we can't begin to fathom how to fathom. It's all based on the assumption that AI will make other better AI. It's not an option, it's an inevitable and logical progression for ML tech.